Method and apparatus for non-model based decentralized adaptive feedforward active vibration control

• US 5,713,438 A
• Filed: 03/25/1996
• Issued: 02/03/1998
• Est. Priority Date: 03/25/1996
• Status: Expired due to Fees

• Current Assignee: Lord Corporation (Parker-Hannifin Corporation)
• Original Assignee: Lord Corporation (Parker-Hannifin Corporation)
• Inventors: Jolly, Mark R., Norris, Mark A., Rossetti, Dino J.
• Primary Examiner(s): Oberleitner, Robert J.
• Assistant Examiner(s): RODRIGUEZ, PAMELA
• Application Number:
  US08/622,540
• Time in Patent Office: 680 Days
• Field of Search: 188/378, 188/379, 188/380, 267/136, 267/140.14, 248/550, 381/71, 244/54, 244/17.27, 364/574
• US Class Current: 188/378
• CPC Class Codes: B60G 2202/25 Dynamic damper
• CPC Class Codes: B60G 2202/42 Electric actuator
• CPC Class Codes: B60G 2400/102 vertical
• CPC Class Codes: B60G 2400/38 Speed of engine rotation
• CPC Class Codes: B60G 2600/1877 Adaptive Control
• CPC Class Codes: F16F 15/02 Suppression of vibrations of non-rotating, e.g. reciprocating systems; Suppression of vibrations of rotating systems by use of members not moving with the rotating systems F16F15/005 takes precedence ; layered products B32B; suppression of vibration in ships B63; relieving load on bearings, using magnetic means F16C39/06
• CPC Class Codes: F16F 7/1011 by electromagnetic means
• CPC Class Codes: F16F 7/108 on plastics springs
• CPC Class Codes: G10K 2210/121 Rotating machines, e.g. engines, turbines, motors; Periodic or quasi-periodic signals in general
• CPC Class Codes: G10K 2210/1291 Anti-Vibration-Control, e.g. reducing vibrations in panels or beams
• CPC Class Codes: G10K 2210/3026 Feedback
• CPC Class Codes: G10K 2210/3027 Feedforward
• CPC Class Codes: G10K 2210/501 Acceleration, e.g. for accelerometers
• CPC Class Codes: G10K 2210/503 Diagnostics; Stability; Alarms; Failsafe
• CPC Class Codes: G10K 2210/511 Narrow band, e.g. implementations for single frequency cancellation
• CPC Class Codes: G10K 2210/512 Wide band, e.g. non-recurring signals

Abstract

A method and system/apparatus implementing a non-model based Decentralized Feedforward Adaptive Algorithm (DFAA) for active vibration control of an actively-driven element, such as an Active Vibration Absorber (AVA) (24). The AVA (24)preferably includes an inertial tuning mass (42) and a voice coil assembly (46) and is contained in an active vibration control system (20) wherein the method and system/apparatus reduce vibration of a vibrating member (22) at an attachment point (26) by receiving an error signal from an error sensor (28) such as an accelerometer and a reference signal from a tachometer (32) or accelerometer (34), where the reference signal is correlated to, or indicative of the frequency content of, a primary vibration source (36) and calculating an updated output signal via an electronic controller (39) using the non-model based DFAA to dynamically drive the actively-driven element, such as AVA (24). The method and system/apparatus using DFAA is effective for reduction of both tonal and broadband vibration. The method approaches the performance of Filtered-x LMS control, yet is decentralized and does not require information regarding the plant.

Claims

1. An active vibration control system for attachment to, and controlling dynamic vibration of, a vibrating member, said dynamic vibration being caused by a primary vibration source, comprising:

(a) at least one actively-driven element for attachment to said vibrating member at an attachment point;

(b) a collocated error sensor for providing an error signal indicative of residual dynamic vibration at a point substantially adjacent to said attachment point on said vibrating member;

(c) a reference sensor for providing a reference signal which is correlated to a frequency content of, said primary vibration source; and

(d) an electronic controller for adaptively calculating an updated output signal which is derived from said reference signal and said error signal based upon a non-model based Decentralized Feedforward Adaptive Algorithm (DFAA) and providing said updated output signal to said at least one actively-driven element thereby causing said residual dynamic vibration at said attachment point to be reduced in one selected from a group consisting of a broadband and tonal fashion.

2. A system of claim 1 wherein said at least one actively-driven element is at least one Active Vibration Absorber (AVA) including;

(i) an inertial tuning mass,

(ii) a flexible element for interconnecting said inertial tuning mass to said vibrating member at an attachment point on said vibrating member, and

(iii) means for actively driving said inertial tuning mass to create a dynamic force acting at said attachment point on said vibrating member.

3. A system of claim 2 wherein said means for actively driving said inertial tuning mass is a voice coil actuator, said collocated error sensor is an accelerometer, and said reference sensor is one selected from a group consisting of a tachometer and an accelerometer.

4. A system of claim 2 wherein said at least one Active Vibration Absorber (AVA) attaches to said vibrating member which is an inner member of an engine mount.

5. A system of claim 1 wherein said non-model based Decentralized Feedforward Adaptive Algorithm (DFAA) is tonal and provides said updated output signal u(k+1) according to an equation:

space="preserve" listing-type="equation">u(k+1)=α

u(k)-μ

b e(k)

where α

is a positive real scalar which affects convergence and steady state solution, u(k) is a complex quantity representative of an output signal to said at least one actively driven element at an operating frequency (fo) at a k^th iteration, μ

is a positive real convergence factor which controls convergence speed, b is a complex constant whose value is dependent upon characteristics of said at least one actively-driven element and said error sensor, and e(k) is a complex quantity representative of a magnitude and phase of said error signal at an operating frequency (fo) at a k^th iteration.

6. A system of claim 1 wherein said non-model based Decentralized Feedforward Adaptive Algorithm (DFAA) is broadband and provides said updated output signal based upon a vector of updated weight coefficients W(k+1) according to an equation:

space="preserve" listing-type="equation">W(k+1)=α

W(k)-μ

e(k) Z(k)

where α

is a positive real scalar which affects convergence and steady state solution, W(k) is a vector of values of filter coefficients for an adaptive filter whose output comprises an output signal to said at least one actively-driven element at a k^th time sample, μ

is a positive real convergence factor which controls convergence speed, e(k) is a complex quantity representative of a magnitude and phase of said error signal at said k^th time sample, and Z(k) is a vector of values derived by passing said reference signal through a B filter dependent upon characteristics of said at least one actively-driven element and said error sensor but independent of system dynamics.

7. A method of reducing dynamic vibration at an attachment point on a vibrating member, said dynamic vibration being generated at said attachment point by a primary vibration source, comprising:

(a) attaching at least one actively-driven element to said attachment point on said vibrating member;

(b) providing an error signal indicative of residual dynamic vibration at a point substantially adjacent to said attachment point from an error sensor which is substantially collocated at said attachment point;

(c) providing a reference signal from a reference sensor which is correlated to a frequency of, said primary vibration source;

(d) calculating in an electronic controller an updated output signal which is derived from said reference signal and said error signal based upon a non-model based Decentralized Feedforward Adaptive Algorithm (DFAA); and

(e) providing said updated output signal to actively drive said at least one actively-driven element and thereby produce dynamic forces which cause said residual dynamic vibration at said attachment point on said vibrating member to be reduced in one selected from a group consisting of a broadband and a tonal fashion.

8. A method of claim 7 wherein said non-model based Decentralized Feedforward Adaptive Algorithm (DFAA) provides tonal control and provides said updated output signal u(k+1) according to an equation:

space="preserve" listing-type="equation">u(k+1)=α

u(k)-μ

b e(k)

where α

is a positive real scalar which affects convergence and steady state solution, u(k) is a complex quantity representative of an output signal to said at least one actively driven element at an operating frequency (fo) at a k^th iteration, μ

is a positive real convergence factor which controls convergence speed, b is a complex constant whose value is dependent upon characteristics of said at least one actively-driven element and said error sensor, and e(k) is a complex quantity representative of a magnitude and phase of said error signal at an operating frequency (fo) at a k^th iteration.

9. A method of claim 7 wherein said non-model based Decentralized Feedforward Adaptive Algorithm (DFAA) is broadband and provides broadband control and provides said updated output signal based upon a vector of updated weight coefficients W(k+1) according to an equation:

space="preserve" listing-type="equation">W(k+1)=α

W(k)-μ

e(k) Z(k)

where α

is a positive real scalar which affects convergence and steady state solution, W(k) is a vector of values of filter coefficients for an adaptive filter whose output comprises an output signal to said at least one actively-driven element at a k^th time sample, μ

is a positive real convergence factor which controls convergence speed, e(k) is a complex quantity representative of a magnitude and phase of said error signal at said k^th time sample, and Z(k) is a vector of values derived by passing said reference signal through a B filter dependent upon characteristics of said at least one actively-driven element and said error sensor but independent of system dynamics.

10. A method of claim 7 wherein said at least one actively-driven element is at least one Active Vibration Absorber (AVA) including:

(i) an inertial tuning mass,

(ii) a flexible element for interconnecting said inertial tuning mass to said vibrating member at an attachment point on said vibrating member, and

(iii) means for actively driving said inertial tuning mass and creating a dynamic force acting at said attachment point on said vibrating member.

11. A method of claim 10 wherein said means for actively driving said inertial tuning mass is a voice coil actuator, said error sensor is an accelerometer, and said reference sensor is one selected from a group consisting of a tachometer and an accelerometer.

12. A method of claim 10 wherein said at least one Active Vibration Absorber (AVA) attaches to said vibrating member which is an inner member of an engine mount.

13. In an active vibration control system including an actively-driven element, an error sensor, and a reference sensor, a control method for reducing dynamic vibration at an attachment point of a vibrating member, comprising the steps of:

(a) obtaining α

which is a positive real scalar between 0 and 1 which affects convergence and steady state solution;

(b) obtaining u(k) which is a complex quantity representative of magnitude and phase of an output signal to said actively-driven element at an operating frequency (fo) at a k^th iteration;

(c) obtaining μ

which represents a positive real convergence factor which controls a convergence speed of said control method;

(d) obtaining b which represents a complex constant whose value is dependent upon characteristics of said driven element and said error sensor in said active vibration control system;

(e) obtaining e(k) which is a complex value representative of magnitude and phase of said error sensor which measures residual vibration at said attachment point substantially collocated with said actively-driven element at said operating frequency (fo) at said k^th iteration;

(f) calculating according to a non-model based Decentralized Feedforward Adaptive Algorithm (DFAA), an updated output signal u(k+1) which is a complex quantity representative of magnitude and phase of said updated output signal at said operating frequency (fo) at a (k+1)^th iteration according to an equation u(k+1)=α

u(k)-μ

b e(k); and

(g) providing said updated output signal to said actively-driven element to drive said actively-driven element and produce active forces which reduce said dynamic vibration at said attachment point in a tonal fashion at said operating frequency (fo).

14. A control method of claim 13 wherein said actively-driven element is at least one Active Vibration Absorber (AVA) including;

(i) an inertial tuning mass,

(ii) a flexible element for interconnecting said inertial tuning mass to said vibrating member at an attachment point on said vibrating member, and

(iii) means for actively driving said inertial tuning mass and creating a dynamic force acting at said attachment point on said vibrating member.

15. A control method of claim 14 wherein said means for actively driving said inertial tuning mass is a voice coil actuator, said error sensor is an accelerometer, and said reference sensor is one selected from a group consisting of a tachometer and an accelerometer.

16. A control method of claim 14 wherein at least one Active Vibration Absorber (AVA) attaches to said vibrating member which is an inner member of an engine mount.

17. In an active vibration control system including an actively-driven element, an error sensor, and a reference sensor, a control method for reducing dynamic vibration at an attachment point on a member, comprising the steps of:

(a) obtaining α

which is a positive real scalar which affects convergence and steady state solution;

(b) obtaining W(k) which is a vector of values of filter coefficients for an adaptive filter whose input is comprised of a reference signal which is correlated to a primary vibration source, and whose output comprises an output signal to said actively-driven element at a k^th time sample;

(c) obtaining μ

which represents a positive real scalar convergence factor which controls a convergence speed of said control method;

(d) obtaining e(k) which is an error signal from said error sensor which measures said dynamic vibration at said attachment point at a point substantially collocated with said actively-driven element at said k^th time sample;

(e) obtaining Z(k) which is a vector quantity and represents an output signal from a filter whose input is said reference signal which is correlated to said primary vibration source and said filter has design parameters which are dependent upon characteristics of said actively-driven element and said error signal;

(f) calculating according to a non-model based Decentralized Feedforward Adaptive Algorithm (DFAA), W(k+1) which is an updated vector of values of filter coefficients for said adaptive filter whose input is comprised of a reference signal which is correlated to said primary vibration source, and whose output comprises an updated output signal to said actively-driven element at a (k+1)^th time sample according to an equation W(k+1)=α

W(k)-μ

e(k) Z(k); and

(g) driving said actively-driven element according to said updated output signal to produce dynamic forces and reduce dynamic vibration at said attachment point within a particular broadband frequency range of said primary vibration source.

18. A control method of claim 17 wherein said actively-driven element is at least one Active Vibration Absorber (AVA) including:

(i) an inertial tuning mass,

(ii) a flexible element for interconnecting said inertial tuning mass to said vibrating member at an attachment point on said vibrating member, and

(iii) means for actively driving said inertial tuning mass and creating a dynamic force acting at said attachment point on said vibrating member.

19. A control method of claim 18 wherein said means for actively driving said inertial tuning mass is a voice coil actuator, said error sensor is an accelerometer, and said reference sensor is one selected from a group consisting of a tachometer and an accelerometer.

20. A control method of claim 18 wherein said at least one Active Vibration Absorber (AVA) attaches to said vibrating member which is an inner member of an engine mount.